Predicting cancer cell invasion by single-cell physical phenotyping

Area of Science:

• Biophysics
• Cell Biology
• Cancer Research

Background:

• Cell physical properties are emerging as critical biomarkers for cancer.
• Distinguishing cancer cell lines and predicting invasion are key challenges in oncology.

Purpose of the Study:

• To identify physical phenotypes that best distinguish human cancer cell lines.
• To establish a predictive model for cancer cell invasion based on physical phenotypes.

Main Methods:

• Quantitative deformability cytometry (q-DC) was used to measure six physical phenotypes at high throughput.
• Machine learning algorithms (k-nearest neighbor and multiple linear regression) were employed for classification and model building.

Main Results:

• Multiparameter analysis of physical phenotypes significantly improved cancer cell line classification accuracy compared to single parameters.
• A set of four physical phenotypes was identified as predictive of cancer cell invasion.
• A validated model demonstrated the ability to predict invasion based on measured physical phenotypes.

Conclusions:

• Physical phenotypes of single cells serve as effective biomarkers for cancer cell classification and invasion prediction.
• Quantitative deformability cytometry (q-DC) combined with machine learning provides a powerful tool for cancer research.
• This approach holds potential for advancing cancer diagnosis and prognosis.